WEBVTT 00:00.890 --> 00:07.010 This is a short wrap up of the period index up ticked so in previous videos we learned that we can either 00:07.010 --> 00:14.630 have a date time index of Earth time stamps or a period index those periods and sometimes that might 00:14.630 --> 00:16.850 be clear up to half a period index. 00:16.850 --> 00:22.780 So for example if we have aggregated monthly data then it might be more appropriate to have periods 00:22.800 --> 00:23.910 labels. 00:24.020 --> 00:28.060 However in this video I want to show you some pitfalls are unexpected. 00:28.050 --> 00:35.240 The behavior of period indexes and that's actually why in most cases I definitely prefer a date time 00:35.240 --> 00:36.350 index. 00:36.350 --> 00:43.700 So first of flag we import pandas and we are working with our temperate set dataset and we are creating 00:43.730 --> 00:46.430 a daytime index so let's have a look here. 00:49.170 --> 00:56.590 So here we have on the left hand side a daytime index with hourly data or hourly increments. 00:56.680 --> 01:02.400 Now let's assume that you want to read sample the data and you want to create a monthly frequency and 01:02.400 --> 01:08.190 we can do this with the risk pool method and actually we can select that we say we want to have a daytime 01:08.260 --> 01:15.240 index by bypassing a timestamp to the prior meter and we take them in and be safe. 01:15.240 --> 01:21.600 Then you aggregated data frame and the variable temperature monthly so let's do this. 01:21.600 --> 01:28.170 So here we have our aggregated data frame with monthly aggregated data and on the left hand side we 01:28.170 --> 01:33.510 can see our daytime and next by default we have fear the very last day of each period. 01:33.510 --> 01:41.160 So the very last day of each month and we can also check again here a 30 inform method and we can also 01:41.160 --> 01:43.900 call here the index attribute. 01:44.280 --> 01:50.250 So at zero day time index and we've already learned that we can index and slice the time series with 01:50.280 --> 01:51.490 a daytime index. 01:51.550 --> 01:53.420 This the cloak operator. 01:53.820 --> 01:56.000 So here for example we want to select. 01:56.040 --> 02:02.150 January 2 13 and we get 1 row of course. 02:02.200 --> 02:13.940 So it's this one year and we can also slice our time series for example from May to 30 into August 213. 02:14.270 --> 02:16.350 And here we get four rows. 02:16.360 --> 02:22.620 So both ends are here actually including and we can simply pass here to a in. 02:22.630 --> 02:30.910 And by doing so we are actually getting all month in the year to 30 and so starting from January into 02:30.940 --> 02:31.640 December. 02:31.930 --> 02:33.940 So in total we I feel twelfth rows. 02:34.600 --> 02:38.080 So this is the behavior when we have a daytime index. 02:38.710 --> 02:49.230 But now let's create a period index by passing year to the kind per meter period. 02:49.560 --> 02:53.700 So again we are aggregating our data from our little monthly. 02:54.840 --> 03:00.840 So I can hear the values in our data frame are exactly the same but we have actually changed to the 03:00.960 --> 03:06.770 next layer of us from one day and the respective periods to the respective periods. 03:06.780 --> 03:10.830 So here we have January February and before we head to one specific day. 03:10.830 --> 03:17.390 So for example the last day or the first day of the period and let's check also you have to the Inform 03:17.390 --> 03:17.820 method. 03:17.850 --> 03:24.840 So we changed you from day time index to period index and we can also confirm this here. 03:26.350 --> 03:33.400 Now let's again assume that we only want to select the very first draw here in January to 13. 03:33.610 --> 03:42.130 So we can do this also with a period index and actually the values here are the same as before but before 03:42.130 --> 03:44.940 we got a data frame now we have a pen not serious. 03:44.940 --> 03:46.560 So no problem here. 03:46.810 --> 03:57.920 And we can also slice our period index from May to August so here we get exactly the same values and 03:57.980 --> 03:59.740 the same structure. 03:59.760 --> 04:05.000 And let's assume that you want to filter for all rows so all periods in the year 230. 04:05.360 --> 04:07.700 So we simply pass the 213. 04:07.700 --> 04:14.040 And now comes the surprise that we only get the very first row into 13. 04:14.090 --> 04:22.710 So here we can see we only get the January 230 and not all 24 hour rows or twelve periods off to 13. 04:22.880 --> 04:30.590 And therefore it might be an option to change to the period index back to a date time index and we can 04:30.590 --> 04:33.410 do this with the method to timestamp. 04:33.680 --> 04:40.840 So we use the method to timestamp on our data frame temperature monthly which has a period index. 04:41.190 --> 04:47.540 And let's try this out here so as to here we have our aggregated monthly data. 04:47.570 --> 04:51.530 But instead of having you on the left hand side a period and next. 04:51.530 --> 04:57.540 Now again we have a date time index and by default we have for the very first day of the period. 04:57.560 --> 05:01.790 So the first of each month and that's the parameter how 05:04.920 --> 05:10.560 so we can define whether we want to have the start of the period or the end and if we can pass either 05:10.560 --> 05:15.260 start R and R S or e so we can also try. 05:15.280 --> 05:15.510 Yeah. 05:15.540 --> 05:15.930 And 05:19.870 --> 05:25.830 until we get actually the very last timestamp of each month and we have actually nominal stacking positions 05:25.840 --> 05:29.050 so that's why we have many nines and uh. 05:29.080 --> 05:31.000 So it's better to have a start 05:34.480 --> 05:35.680 and actually as a summary. 05:35.680 --> 05:42.880 So this is my personal taste that as far as I can I definitely stick to a day time index. 05:42.970 --> 05:49.960 So I feel comfortable or more comfortable with a day time index but as I said this is a matter of taste. 05:49.960 --> 05:54.340 And if you're like You can also dig deeper into the period index. 05:54.430 --> 05:56.720 So feel free to do this. 05:56.890 --> 06:02.350 And with this we are finished with this video and I hope to see you also in the next one by.